digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id
85 I S M O S A T, Proceeding International Symposium For Modern School Development, ....
lems and has already achieved the best value, this value is called p best. Value of the “best”
others is the best value achieved by any par- ticle in the population, this value is called g
best. PSO has avelocity that would change the position of particles on each iteration. At
each iteration the value of the velocity and position of renewed[12].
Equation of PSO algorithm consists of velocity and position, the most fundamental
of which is as follows, velocity:
++
. :5;:;G
. :5;
21 H1
H1 2
1 2 . 2
.
2 1 2 1I.
1
21. 2
+ .
J :; :; . .
. =
+ B
B K LL
+ =
:; B +
2 2
2 2
2 2
, 2
2
:A; - 2
+ +
M .
+ D
. 5 .
:;
N :;
where: i
= particle index k
= iteration v
= velocity of particle x
= positionof particle p
= the best position of the particle pbest G
= the best position of the swarm gbest L
1,2
= learning rates R
1,2
= random numbers with interval [0 – 1]
W = inertia
In the method of standard PSO imple- mentation, it was found that the velocity of
particles in PSO updated too fast and the minimum value of the objective function is
often over looked. There fore, there is a re- vision or improvement of the standard PSO
algorithm. Improvements in the form of the addition of an inertia
θ to reduce speed. Usu- ally the value of
θ is made such that increas- ing iterations passed, the smaller the particle
velocity. This value varies linearly within the range of0.9to0.4. This inertia weights used to
dampen the paceduring the iterations, which allows birds to the target point more accu-
rately and eficiently than the original algo- rithm [9]. High inertia weight values increase
the portion of the global search global ex- ploration, whilea low value emphasizes lo-
cal search local search. For not very focused on one part and keep looking for new search
area in particular dimensional space, it is necessary tobe sought inertia weight value
θ which draw maintaining global and local search and to reach it and accelerate conver-
gence, aweight of inertia that decreases in value with increasing iterations used by the
formula [9]:
Figure2. Model of the optimal control system on temperature regulation
digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id
86
5 -
K -
K 9
11
1 ,
.
9
J+ =8
4 =8
8 + B
O B
-9 + B +
- .
= ,94 K
8. J
3J=8J44 J
= 9+B. P55 CJ 9+B.
K8
8, 9+B.
-9
G 9+B.
where: W
max
: Value of the initial weight W
min
: Value of the inal weight It
: Iteration running It
max
: Maximum iterations
In this paper, the variables are optimized as much as 3 variables to determine the
parameters of PID control parameters in the form constants Kp, Kd and Ki.
4. Temperature Control System
The structure of the optimal control sys- tem on temperature regulation built in Simu-
link matlab application program, as igure 2. PID controlis packaged in a subsystem
PID-PSO with the input set point and feed- back ADC. Data from the temperature sensor
in the form of a voltage serially connected to arduino with baudrate 115200 through com-
port 1 and then converted into scale ADC to be used as a comparative calculation so that
it can display temperature reading sin unitso C. PSO sub system contains the model PID
combined with 3 gain to be optimized. Gain
PID constants contained in the Error and Δ Error input and PWM output. On Error given
scope with the data stored in the work space as the reference data at the time of optimiza-
tion by PSO.
5. Simulation of Temperature Control System
The design of the electronic device in- cludes a temperature sensor, heater blower
and Arduino Rev.3. Sensors using LM35 with output voltage of 10 mV per 1
o
C. To further improve the accuracy and precision on tem-
perature measurements, carried out set-up voltage reference 8-bit ADC with a range of
0-255 represent the minimum and maximum temperature can be measured by the circuit.
Sensor output connected into non-inverting
ampliier circuit to improve the precision of measurement. Actuator circuit using a MOS-
FET IRFZ44 N-type MOSFETs supplied with 12 VDC according to the needs of the blower
and DC Lamp that will be controlled. MOS- FET is controlled by the PWM signal from ar-
duino to determine the value of the incoming voltage to the circuit. Blower connected in
series with a MOSFET as electronic switches. Figure 3 is a 3D model of the hardware used
in the simulation process.
Figure 3. 3D model of the hardware simulator
Optimization process is then performed using PSO with LDIW methods to obtain pa-
rameters optimal control system, optimiza- tion results achieved convergence on the 11th
iteration.
+
+ ,-
5+,-
5 .
+ ,- 1 2 45 1 2 5
1 2
45 .=B
. B 2 456
4
5Q8
4 J+,-
. 5
63 46 4
R
S 8-
B
37 +
+ ,- B OB
-9
. +
,-
. 8 + . 01,
01,22013 4
4 5
6 665 7 63 46 37
- +
,94 =8
9 J + -9
+ 5 8
37 637
J+
7
.
+ ,- 63
46 J+
6 665
:; =
T= =
1 D
=
B
. G 9 U DCS
:; =GT + G
9 = 344U 1 V . D
?5 :;
G T.
8 ,U 50A?5
:5; 1
= 8
S , T U
5 44A46? 4
:4;
C T9 K ,
8 U
=.B8 =
.
= . :; B
? 1
T=
U 4A5
:6;
C T9
+ K
8 + G - D ,8 =
8 .
G+U
Figure 4. itness convergence of iterations PSO with LDIW.
Figure 4 shows that the convergence of particles occurred on the 11th iteration. The
resulting constant value of PSO with LDIW method is stored in the workspace, where
the optimal parameter values produced are: KP = 9.5401, KI = 4,666 and KD = 0.9504 with
a value ITAE Integrated Time absulute Er- ror = 6506.4971. Furthermore, the PID con-
trol simulation using the optimal parameters with sampling time 150 and the setpoint tem-
perature 40 ° C, following the curve of simu- lation results.
digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id digilib.uinsby.ac.id
87 I S M O S A T, Proceeding International Symposium For Modern School Development, ....
Figure 5. Rise time and Settle time of the simulation results PID-PSO with LDIW.
The simulation results show that the control system response curves correspond-
ing to the set point 40
o
. Rise time is reached at the 7:18 time simulation and settle time at
15.76 time simulation, as in Figure 5, where the X axis is time simulation and the Y axis is
in units of temperature
o
C.When the setpoint is reached and the control system running,
Error generated 0:18. Overshoot does not occur when raise the target because the PSO
with LDIW better methods for changing the
value of the gain Error, ΔError and output so PWM value given is not too high. The follow-
ing comparison table PSO simulation results without and with LDIW.
Table 1. Comparison of the simulation results PSO
Setpoint
PSO ixed W = 0.4 PSO with LDIW
0.4W0.9 Rise
time Settle
time Error
Rise ime
Settle time
Error 40
7.63 27.74
0.26 7.18 15.76
0.18
6. Conclusion
PID controller based on PSO can be applied to a prototype plant temperature
control with LM35 temperature sensor as a feedback control system through a port con-
nected arduino serial ADC integrated with Matlab-Simulink. Optimization results ex-
pelled through PWM port arduino to control
the heater and blower, so that the deined set point temperature can be achieved steady
state. On set point temperature of 40
o
C, the average value obtained error 0.18, com-
pared with manual PID tuning is 0.78 and the PID-PSO ixed inertia 12.26.
The use of PID controller based on PSO with LDIW method shows faster perfor-
mance, with rise time 7.18 ms, settle time 15.76 ms compared with the results of the
PID control - PSO ixed Inertia, rise time is reached at 7.63 ms, settle time 27.74 ms.
References
[1] A. Hasyim, “Alat Pengatur Suhu Air
dan Keluaran Udara Sebagai Penggerak Air Pada Ember Penetas Telur Ikan
Gurameh Berbasis Mikrokontroler,” Universitas Negeri Yogyakarta, 2013.
[2] A. Binsar Parmonangan, “Pengatur
Suhu Ruangan Otomatis Dengan Sensor Suhu Berbasis Mikrokontroler
Atmega 8535,” Fakultas Ilmu Komputer Teknologi Informasi, Universitas
Gunadarma, Jakarta, 2014.
[3] P. Singhala, D.. Shah, and B. Patel,
“Temperature Control using Fuzzy Logic,” Int. J. Instrum. Control Syst., vol.
4; 1, pp. 1–10, Jan. 2014.
[4] K. H. Ang, G. Chong, and Y. Li, “PID
control system analysis, design, and technology,” IEEE Trans. Control Syst.
Technol.
, vol. 13, no. 4, pp. 559–576, Jul. 2005.
[5] H. Fachrudin, I. Robandi, and N.
Sutantra, “Model and Simulation of Vehicle Lateral Stability Control,”
presented at the 2nd APTECS, 2010, International Seminar on Applied
Technology,
Science, and
Arts, Surabaya, ITS, 2010, p. 26.
[6] R. Eberhart and J. Kennedy, “A new
optimizer using particle swarm theory,” in, Proceedings of the Sixth International
Symposium on Micro Machine and Human Science, 1995. MHS ’95
, 1995, pp. 39–43. [7]
H. Fachrudin, I. Robandi, and N. Sutantra, “Modeling and Simulation
